Splenic Marginal Zone Granulocytes Acquire an Accentuated Neutrophil B-Cell Helper Phenotype in Chronic Lymphocytic Leukemia

Chronic lymphocytic leukemia (CLL) is a low-grade B-cell tumor characterized by an accumulation of monoclonal CD5⁺ mature B cells in secondary lymphoid organs (SLO), bone marrow, and peripheral blood (1). Gene expression profiling (GEP) showed that these malignant B cells resemble antigen-experienced memory B cells (2, 3). Although many circulating CLL cells are in a quiescent G₀–G₁ cell-cycle phase, their proliferative capacity can resume provided that they get access to stimuli within specific niches (4). The concept of the microenvironment as a pivotal regulator of CLL antiapoptotic signaling and cell division implies a leading role of antigen stimulation through the B-cell receptor (BCR; ref. 4). Accordingly, inhibition of Bruton tyrosine kinase led to an impressive clinical success (5). Accessory signals provided by the microenvironment, that is, those perceived by Toll-like receptors (TLR; ref. 6), and B-cell growth factor receptors (7), are concomitantly engaged with the BCR. In addition, a strong influence of excess growth signals from the microenvironment, including B-cell–activating factor (BAFF), could already be involved in clonal selection and costimulate putatively autoreactive B cells (8).

In vitro, CLL cells rapidly undergo apoptosis from which they can be rescued by coculturing with stromal cells and cytokines (9). This interaction provides tumor cells with prosurvival signals and conversely, stromal cells are transformed toward a tumor-permissive niche. In vitro, a hallmark of lymphoma-induced stroma polarization is the activation of the NF-κB pathway associated with an inflammatory lymphoid stroma phenotype (10, 11).

Recently, we have shown that murine CLL cells in the transgenic Eμ-Tcl1 model migrated CXCR5 dependently into B-cell follicles where they lodged adjacent to follicular dendritic cells (FDC). Leukemia B cells trafficked in a marginal zone (MZ) B-cell–like manner as they moved directly from the MZ sinus across the MZ-white pulp border into the B-cell follicle (12). It remains still unclear how tumor cells receive antigenic stimulation and second, what role accessory signals from the microenvironment have on top of BCR activation.

Here, we defined tumor stroma as the nonmalignant tumor cell microenvironment, including mesenchymal cells, vasculature, extracellular matrix, but also hematopoietic cells. Functionally, stroma forms a growth and survival niche for lymphoma B cells in SLOs. We analyzed the leukemia cell–imposed stromal alterations along their trafficking route including the splenic MZ, the B-cell follicle, and the FDC network. Adoptive tumor cell transfers or in vitro cell cultures are often skewed towards a strong inflammatory component and may mimic rather a tumor initiation phase. We performed GEP on laser capture microdissected tissues obtained from spontaneously diseased Eμ-Tcl1 mice, a condition that might better reflect an equilibrium or escape phase in tumor progression (13, 14). We demonstrate a splenic tissue remodeling predominated by the expansion and polarized differentiation of monocytic cells and neutrophils. This innate immune cell infiltration was not associated with a systemic inflammation. In established CLL, tumor-associated macrophages (TAM) exhibited an M2 polarization, whereas tumor-associated neutrophils (TAN) in spleen were skewed toward a neutrophil B-cell helper–like (N_BH) phenotype. Selective depletion of myeloid subpopulations in mice retarded leukemia progression substantially. Innate immune cells provided additional growth factors, but were also involved in the structural integrity of the spleen necessary for leukemia cell trafficking and homing.

A comprehensive list of all antibodies is given in Supplementary data.

Eμ-Tcl1 transgenic mice on a mixed C3H/C57BL/6 (B6C3F1) or C57BL/6 background (F10) were used as described previously (12). C57BL/6-Gt (ROSA)26Sor<tm1(HBEGF)Awai>/J (iDTR) mice were bred to LysM-Cre mice to create LysM-Cre/iDTR mice where the diphtheria toxin receptor (DTR) is expressed in macrophages. MARCO gene-deleted mice (MARCO^−/−) were used as described previously (15). An extended description of mice is given in Supplementary data.

Cells were blocked with CD16/32 antibody and further antibody stained and washed in FACS buffer [PBS, 0.5% w/v BSA, 0.05% (v/v) NaN₃]. A flow cytometry gating strategy for leukemia cells was performed as described previously (12). Neutrophils were gated based on SSC/FSC scatter plot, and CD11b⁺Ly6G⁺ reactivity. Macrophages were gated by CD11b⁺Ly6C⁺ or CD11b⁺F4/80⁺ staining. Data were acquired on a FACSCantoII flow cytometer and analyzed with FlowJo software (TreeStar). Cell sorting was done on a FACSAria (BD Biosciences).

Neutrophils from femurs and tibias were flushed with PBS and isolated by Percoll density gradient centrifugation. Superoxide was measured using the superoxide dismutase-inhibitable ferricytochrome C reduction assay as described previously (16). Generation of intracellular reactive oxygen species was measured by oxidation of CM-H₂DCFDA (Sigma-Aldrich). Bone marrow or splenic neutrophils (1 × 10⁷/mL HBSS) were loaded with 1 μmol/L CM-H₂DCFDA for 15 minutes at 37°C. A total of 2.5 × 10⁵ cells were incubated with buffer control, PMA, or opsonized zymosan. Reactions were stopped after 60 minutes by adding PBS/1% BSA. The shift of green fluorescence in Gr-1⁺ neutrophils was determined and the mean fluorescence intensity (MFI) is reported.

Gene expression profiling was done as described previously (12). An extended description of the method is given in Supplementary Data.

Gene set enrichment analysis (GSEA) was performed as described previously (17) against an integrated database containing the Molecular Signature Database v3.1 (18), the GeneSigDB (19) and the Staudt Lab library (20). Signatures of human genes were translated via gene homology. Only gene signatures that displayed a significant enrichment [P < 0.005, false discovery rate (FDR) < 5%] as well as those that contained at least 20% significantly differentially expressed genes (P < 0.05 by two-sample t tests when comparing leukemia B cells with Wt stroma samples, respectively, by paired t tests when comparing measured with hypothetical mixtures) were considered to represent differentially regulated pathways. Signatures with less than 10 genes were filtered out.

Recipient mice were injected intravenously with 1–2 × 10⁶ Eμ-Tcl1 cells. For the depletion of neutrophils, the anti-Ly6G clone 1A8 (Biolegend) or a rat IgG2a k isotype control antibody were injected intraperitoneally.

Bromodeoxyuridine (BrdUrd) incorporation into proliferating leukemia cells was analyzed as described previously (12). BrdUrd was added at 10 μmol/L daily into cocultures of leukemia cells and isolated neutrophils (cell ratio of 1:10). For BAFF blockade, a recombinant mouse BAFF receptor/Fc chimera (BAFFR-Ig) was used (R&D Systems), and for APRIL neutralization an anti-APRIL antibody (clone: Apry-1-1; Adipogen) was added.

Apoptosis of splenic CD5⁺CD19⁺ leukemia B cells was assessed with flow cytometry applying AnnexinV–FITC and propidium iodine (PI) staining (eBioscience) exactly as described by the manufacturer.

Eμ-Tcl1 mice were treated over 8 days with three injections intraperitoneally of an anti-IL10 (500 μg; clone JES5-2A5, Biolegend) and anti TGF-β (500 μg; clone 1D11.16.8, BioXCell) antibody. Alternatively, an anti-IL10 receptor (IL10R) antibody (clone 1B1.3A, BioXCell) was applied. For mock control, a rat IgG1 isotype antibody was used.

Results are expressed as arithmetic means ± SEM if not otherwise indicated. Values of P < 0.05 were considered statistically significant, as determined by the Mann–Whitney U test, the unpaired or paired Student t test, or the Wilcoxon signed rank test where appropriate.

Using genome-wide expression arrays, we examined relative changes in gene expression from naïve wild-type (Wt) stroma compared with tumor-bearing Eμ-Tcl1 stroma covering B-cell follicles and adjacent MZ. GEPs from sorted CD5⁺CD19⁺ tumor cells were determined separately (12). We utilized a linear biostatistical mixture model whereby tumor cell–induced changes in gene expressions of stroma cells could be identified selectively. We identified 502 genes that were selectively upregulated (P < 0.01) in tumor-bearing stroma compared with naïve Wt mice (Fig. 1A). The Venn diagram illustrates the specificity of the biostatistical model to detect changes solely in tumor-bearing stroma because only 55 overlapping genes between both analysis groups were detected.

To gain insights into biologic processes, we used the 126 top upregulated genes (P < 0.001; Supplementary Table S1) as input for Gene Ontology (GO) analysis. Significant overrepresentation of the terms neutrophil chemotaxis (GO:0030593, 7/33 genes are among the top 126 upregulated genes of all 21,225 measured genes, P = 8.3e−10, hypergeometric test), innate immune response (GO:0045087, 11/176 genes, P = 8.3e−9), and neutrophil aggregation (GO:0070488, 2/2 genes, P = 3.5e−5) was obtained. The top 25 upregulated genes [log₂(ratio) ≥ 1.5] contained markers that are preferentially expressed by neutrophils [e.g., Mpo, log₂(ratio) = 1.6, P = 0.002; Ly6G, log₂(ratio) = 1.9, P = 1.4e−5], DCs [e.g., CD209b, log₂(ratio) = 2.9, P = 5.9e−6], and macrophages [e.g., Marco, log₂(ratio) = 1.8, P = 2.2e−5; Fig. 1B; Supplementary Table S2).

GSEA confirmed enriched signatures related to granulocytes, innate immune response, DCs, macrophages, and monocytes (Supplementary Fig. S1A–S1J). An exemplary heatmap depicting the signature "LIAN_LIPA_Targets_3M" contains genes indicative of granulocyte (e.g., Cxcr2, Ccr1, C3ar1, CD244), macrophage (CD68, Marco, Mafb), and DC-associated (Tlr7, Spic) expression (enrichment score -0.77, P = 0.001 via permutation test, FDR = 0.001; Fig. 1C and D).

Flow cytometry analysis of myeloid leukocytes was aimed at a discrimination of differentiation states depending on the anatomic context. We found enhanced frequencies of myeloid cells (CD11b⁺) in leukemia-bearing Eμ-Tcl1 mice compared with Wt mice in spleen, bone marrow, or peripheral blood (Fig. 2A and B). Total numbers of CD11b⁺ cells were increased 2.5-fold in spleen, but not in bone marrow. An increase in frequencies and numbers (5.5-fold) of inflammatory monocytes (CD11b⁺Ly6C^high) as well as for patrolling monocytes/macrophages (CD11b⁺Ly6C^int; 4.9-fold) was obtained in spleens from Eμ-Tcl1 mice (Fig. 2C and D).

Next, we compared gene expression from in vitro differentiated M1 or M2 human monocytes with genes upregulated in Eμ-Tcl1 stroma (21). Applying the Martinez and colleagues gene signature, in tumor-exposed splenic stroma, a significant enrichment of an M2 signature (enrichment score = −0.61, GSEA P ≤ 0.001 by permutation test and P = 0.001 by t test between signature averages), was calculated (Supplementary Fig. S2).

In spleens, bone marrow, and peripheral blood from Eμ-Tcl1 mice, an expansion in neutrophil (CD11b⁺Ly6G⁺) numbers (4.6-fold) and frequencies occurred (Fig. 2E). In serum the release of proinflammatory mediators (Fig. 2F) was indistinguishable between Wt and Eμ-Tcl1 mice. GM-CSF and the proinflammatory marker S100A8 remained below the detection limit. Hence, increased numbers of innate immune cells were not associated with a systemic inflammation.

To assess the conditions for stroma remodeling, we analyzed "secreted protein acidic and rich in cysteine" (SPARC) gene expression. Under autoinflammatory settings, lack of SPARC resulted in a disturbed compartmentalization within SLOs (22). Here, SPARC was weakly upregulated in Eμ-Tcl1 spleens compared with controls [log₂(ratio) = 0.7; P = 0.0053]. Collectively, the innate immune response signature of Eμ-Tcl1 mice was characterized by an expansion of granulocytes and monocyte subsets, but lacked serum markers for an acute inflammatory state.

An expanded neutrophil population (Ly6G⁺) was seen adjacent to the MZ (MARCO⁺) and in the red pulp of Eμ-Tcl1 mice. Neutrophils were absent from the B-cell follicle, as marked by the metallophilic marginal macrophage marker MOMA-1⁺ and B220⁺ staining (Fig. 2G). Next, we chose human splenic marginal zone lymphoma (SMZL) to compare localization of granulocytes with the Eμ-Tcl1 model. SMZL and murine CLL B cells are both localized in the splenic MZ, grow in an indolent manner, and share a strong dependency on BCR and TLR signaling (23). In SMZL compared with normal spleen, we observed a stronger neutrophil infiltration (MPO⁺) into the neoplastic MZ, a reduction of the MZ-red pulp (RP) demarcation, and a tight intermingling between neutrophils and lymphoma B cells (Supplementary Fig. S3). Thus, the murine model mimics microanatomic aspects of granulocyte infiltration into distinct splenic compartments as seen in an indolent human lymphoma.

The macrophage receptor with collagenous structure (MARCO; Supplementary Table S2) confers MZ macrophages with a high capacity for particle uptake important for MZ B-cell interactions. Because Eμ-Tcl1 leukemia cells traffic in a MZ B-cell–like manner, we explored the clinical disease course in MARCO^−/− mice after adoptive cell transfer. Up to 32 days, tumor load in MARCO^−/− recipients was substantially lower compared with Wt animals (Fig. 3A). To deplete additional macrophage subsets, we used the Lysozyme M–directed myeloid-specific DT (LysM-Cre/iDTR) transgenic mouse strain. After DT or NaCl administration, we adoptively transferred Eμ-Tcl1 leukemia B cells into LysM-Cre/iDTR recipients. Macrophage depletion retarded substantially tumor progression in the spleen (Fig. 3B). A 3.4-fold reduction of the spleen-resident F4/80⁺CD11b⁺ macrophage population was seen in DT-treated mice. Reduction of MOMA-1⁺ and MARCO⁺ MZ macrophages in DT-treated LysM-Cre/iDTR mice could be confirmed (Fig. 3C).

Next, we explored the possibility that splenic MZ macrophages control leukemia cell translocation into B cell follicles. Labeled tumor cells were injected into LysM-Cre/iDTR mice pretreated with DT or NaCl. Five hours later, spleens were stained for B-cell follicles, and with anti MAdCAM-1 to highlight the border between the MZ and the follicles (Fig. 3D). We quantitated the numbers of labeled leukemia cells within the follicles (B220⁺) and outside in MZ (MAdCAM-1⁺/B220⁻) and RP. A substantial reduction by 39% in the total number of leukemia cells in the RP and MZ of DT-treated LysM-Cre/iDTR compared with mice with a preserved MZ macrophage ring was observed (Fig. 3E). At this early time point, only a small proportion of tumor cells was detectable in the follicles. Collectively, this indicated that macrophages regulate the retention of Eμ-Tcl1 leukemia cells in the MZ.

Neutrophilic granulocytes constitute a significant fraction of inflammatory cell infiltrates found in many solid cancers. They are also referred to as TANs and depending on their activation and differentiation state, anti- and protumorigenic functions have been proposed (24).

GSEA revealed an enrichment of upregulated genes from signatures coding for granulocytes (enrichment score = −0.706, GSEA P = 0.001 by permutation test; P = 4e−5 by t test between signature averages; Fig. 4A). We confirmed the conditions for neutrophil recruitment into leukemic spleens as indicated by the significant overrepresentation of the GO term "neutrophil chemotaxis" covering chemotactic factors (S100A8, S100A9), receptors (CXCR2), inflammatory cytokines (IL1b), and integrins (Itgam/CD11b; Fig. 4B).

We depleted neutrophils using a Ly6G-specific antibody over four weeks. Eμ-Tcl1 mice treated with the isotype antibody developed a 2-fold higher splenic tumor load compared with anti-Ly6G–treated animals (Fig. 4C and D). Disappearance of CD11b⁺Ly6G⁺ cells after anti-Ly6G-treatment was confirmed by flow cytometry and immunohistology (Fig. 4C and Supplementary Fig. S4A). In control spleens, neutrophils were found predominantly in the RP, but also in the MZ where they tightly intermingled with leukemia B cells (SNARF-1⁺). Neutrophils remained excluded from the T cell and follicular B-cell zone.

Splenic-resident leukemia B cells from neutrophil-depleted mice had a lower proportion of cells in S-phase (BrdUrd⁺ isotype-treated, 10.72% ± 0.6; Ly6G-treated 6.8% ± 0.4), but no change in the apoptosis rate (Fig. 4E). Granulocyte depletion did not cause major changes in the inflammatory cytokine profile (Supplementary Fig. S4B).

Next, we performed cocultures of isolated splenic neutrophils and Eμ-Tcl1 tumor cells. Neutrophils taken from either Wt or Eμ-Tcl1 mice enhanced the BrdUrd uptake in leukemia cells about 2-fold (Fig. 4F). However, we noted that splenic neutrophils in Eμ-Tcl1 mice outnumbered those obtained from Wt mice substantially (Fig. 2E).

Growth promoting and antiapoptotic stimuli for MZ B cells are often regulated via the NF-κB transcriptional pathway. When neutrophils were depleted, an obvious change of NF-kB target genes in leukemia cells was not seen. In the neutrophil-depleted group the transcription factors early growth response 1 (Egr1) and Egr2 were significantly upregulated 1.7-fold (Egr1) and 1.6-fold (Egr2; Fig. 4G). Egr1 enhances p21^Waf/Cip1 promoter activity, which results in cell-cycle arrest at the G₂–M phase. At high expression Egr1 maintains hematopoietic stem cell quiescence and counteracts their migratory capacity. Egr2 seems to be crucial for defined B-cell developmental steps (25, 26). Collectively, Egr1 and Egr2 transcriptional activities could control leukemia B-cell activation, depending on the crosstalk with neutrophils.

Under physiologic conditions, a spleen-resident N_BH population that receives reprograming signals from sinusoidal endothelial cells exists (27). TANs had a 1.5-fold higher gene expression for Baff and a 3-fold higher April expression compared with Wt mice (Fig. 5A). Because BAFF concentrations in serum were not different (Fig. 5B), we isolated splenic neutrophils and stimulated them with PMA. Induced secretion of BAFF was 2.4-fold higher in neutrophils from leukemic animals (Fig. 5C).

Next, we isolated granulocytes from bone marrow and cocultured them with Eμ-Tcl1 tumor cells for 48 hours. In the presence of an inhibitory BAFF-receptor fusion protein in combination with an anti-APRIL antibody, BrdUrd uptake in tumor cells was reduced (Fig. 5D).

Gene expression for several chemokines (Ccl3, Ccl4, Ccl5, Ccl6) and chemokine receptors (Ccr1, Ccr3, Cxcr2, Cxcr3) was downregulated (Supplementary Fig. S5) in tumor-challenged animals. Other inflammatory cytokines (IL1, Ltβ) and receptors (IL6ra) were downregulated as well. Notably, the same cytokine and chemokine gene markers were indistinguishable when analyzed in bone marrow–derived neutrophils, indicating that microenvironmental conditions can lead to a distinct neutrophil differentiation. Splenic neutrophils were cultured for 24 hours in the presence of G-CSF. The rate of apoptotic neutrophils from Eμ-Tcl1 mice was 2.4-fold lower compared with Wt mice, as revealed by a higher frequency of AnnexinV⁻/PI⁻ cells (Fig. 5E).

We isolated neutrophils from bone marrow and used PMA and zymosan that stimulate the respiratory burst via different signaling pathways. Release of the reactive oxygen (ROS) species O₂⁻ was indistinguishable (Fig. 5F). Likewise, H₂O₂-catalyzed hydrolysis of the dye CM-H₂DCGFDA into the fluorescent dye DCF was similar between Wt and Eμ-Tcl1 neutrophils (Fig. 5G).

The cytokines IL10 and TGFβ to contribute to splenic neutrophil reprograming (24, 27). In GEP, Tgfβ gene expression itself was unchanged in leukemic stroma (Tgfb1–3, P > 0.05); however, a strong upregulation of the TGF-β–induced gene Tgfbi was seen [log₂(ratio) = 1.3, P = 0.005]. This gene encodes an extracellular matrix protein that can bind integrins and thus, confers adhesion to monocytes (28). In tumor-bearing mice, TGFBi was much stronger associated with MAdCAM-1⁺ structures in the B-cell follicle and MZ (Supplementary Fig. S6). For IL10, we found a strong gene upregulation in leukemia cells relative to Wt stroma [log₂(ratio) = 2.97, P = 5e−5], MZ B-2 [log₂(ratio) = 2.99] or B-1 [log₂(ratio) = 2.52] lymphocytes. Release of IL10 from leukemia cells was 39-fold higher compared with Wt B cells. In serum from Eμ-Tcl1 mice, the immunoinhibitory cytokine was increased (mean Wt: <5 pg/mL, Eμ-Tcl1: 693 pg/mL; Fig. 6A and B). Tumor-bearing animals were treated with neutralizing anti-IL10/anti-TGFβ antibodies. Downregulation of CD62L and upregulation of CD16/32 indicated that splenic neutrophils maintained their activation level in an IL10 and TGFβ dependent manner (Fig. 6C). LAIR-1 as the main inhibitory receptor expressed on activated neutrophils remained unaltered.

Next, we compared gene expression of sorted neutrophils (CD11b⁺Ly6G⁺) from Wt and Eμ-Tcl1 mice. The gene markers Ikbke, Bcl2a1a, and Mpo were chosen because they discriminate between granulocytes and granulocytic-myeloid derived suppressor cells (G-MDSC; ref. 29). Neutrophils from Eμ-Tcl1 mice exhibited lower gene expression for Mpo (Fig. 6D), indicating a mature differentiation state. Expression of the antiapoptotic gene Bcl2a1a was 2-fold higher in neutrophils from Eμ-Tcl1 mice (Fig. 6D).

To corroborate that IL10 and TGFβ led to neutrophil reprograming and subsequently, leukemia expansion, we treated Eμ-Tcl1 mice with a combination of anti-IL10R and anti-TGFβ antibodies. In the anti-IL10R/anti-TGFβ treatment group, the transcription factor Egr1 was upregulated in isolated Eμ-Tcl1 cells, indicating that tumor cells had entered cell-cycle arrest. This result was consistent with the effects seen when neutrophils were depleted (see Fig. 4G). Bcl2a1a, Baff, and April gene expressions in neutrophils showed a tendency to fall when animals were treated with anti-IL10R/anti-TGFβ antibodies.

Collectively, IL10 and TGFβ contribute to neutrophil differentiation in Eμ-Tcl1 mice.

We used GEP to identify the stromal alterations that occur when an oncogene-driven lymphoid neoplasia lodges within murine spleens. In the absence of an overt proinflammatory cytokine milieu, we showed that macrophages and neutrophilic granulocytes cooperatively enhance the progression of CLL in mice.

In human follicular lymphoma, the adverse immune-response signature 2 included genes associated with monocytes and DCs (30). In Eμ-Tcl1 mice, we validated expanded monocyte and neutrophilic granulocyte populations, a situation that pointed to a link between inflammation and cancer (13). Increased numbers or an altered differentiation state of innate immune cells is commonly observed in human chronic inflammation, which bears a substantial risk for extranodal marginal zone B-cell lymphomas (31).

To further define the relationship between innate immune cell infiltration and a putative sterile inflammation in vivo, we assessed proinflammatory cytokines in serum from diseased mice; however, these mediators were similar to controls. Recently, it was suggested that anti PD-L1 blockade reverses a CLL-induced skewing of myeloid cells, which was associated with the reduction of selected inflammatory cytokines (32). These effects were seen after adoptive tumor cell transfer and are therefore not comparable with our results from transgenic mice.

By unbiased GSEA, we found no obvious enrichment for transcriptomic signatures dependent on the proinflammatory transcription factors NF-κB, AP-1, STATs, CEBP/β, and HIF-1α (33). In vitro, mesenchymal stromal cells (MSC) obtained from follicular lymphoma–infiltrated bone marrow exhibited a higher CCL2 expression leading to recruitment of monocytes, followed by their polarization into a TAM-like phenotype (34). Bone marrow–MSCs produced more neutrophil-attractant IL8 and were activated via TNF and NF-κB signaling pathways (11). In contrast, we here analyzed stroma gene signatures under in vivo settings, which might better reflect systemic conditions of cellular networks. The occurrence of an M2-skewed phenotype of TAMs indicated the gradual inhibition of NF-κB activity and is consistent with M2 polarization found in established tumors. On the contrary, activated NF-κB in myeloid cells is rather associated with tumor promotion in inflammation-associated cancer models (35).

In contrast to M1 inflammatory macrophages, macrophages that occur in cancer tissues are polarized toward an M2 phenotype (36). These cells enhance tumor progression, either through direct secretion of cytokines that sustain tumor cell proliferation, or by dampening the immune response through secretion of immunosuppressive cytokines and overexpression of T-cell–inhibitory ligands (14, 32, 37–39). In CLL, monocyte-derived nurse-like cells exist that resemble M2-polarized TAMs and support leukemia cell survival in vitro (40, 41). In diffuse large B-cell lymphoma a high density of an M2-polarized TAM population correlated with an unfavorable prognosis (42).

The functional relevance of macrophages in human lymphoma is still controversial (43–45). These uncertainties relate to their lineage plasticity, which could be influenced by EBV infection and cytokines, but also by the microanatomic localization (46). Here, we obtained a substantial increase of inflammatory Ly6C^high monocytes and, on the other hand, patrolling Ly6C^int monocytes were expanded as well. In a liver model of sterile inflammation a phenotypic conversion between Ly6C^high and Ly6C^int occurred (47). This functional plasticity was controlled by IL4 and IL10, which implies that Eμ-Tcl1 leukemia cells themselves might provide the stimuli for M2 macrophage differentiation (13).

Macrophages in the spleen are strategically located to recognize and retain blood-borne antigens (48). We observed delayed Eμ-Tcl1 leukemia progression upon genetic deletion of the scavenger receptor MARCO and of MZ and metallophilic macrophages (15). These results indicated that leukemia cells do not only traffic in a MZ B-cell–like pattern, but may cooperate with local macrophage subsets in a manner similar to benign MZ B cells. This interaction regulates antigen uptake by macrophages and antigen capture by B cells, followed by B-cell trafficking into the follicle and antigen delivery onto FDCs (49). When splenic macrophages were present, leukemia cells accumulated stronger in the MZ and adjacent areas as compared with macrophage-depleted animals. Our finding is consistent with the dynamics of physiologic B lymphocyte distribution in macrophage-depleted animals (50). In this scenario, MZ macrophages might control CXCR5/CXCL13-guided leukemia B-cell trafficking into B-cell follicles where a stronger FDC-dependent growth stimulus becomes accessible (12).

Neutrophils exhibit substantial plasticity depending on environmental stimuli (51). Under the conditions of a tumor-prevalent TGFβ milieu the differentiation of a tumor-supporting N2 polarization can occur (24). In agreement with an altered differentiation depending on early versus late tumor stages, splenic neutrophils in Eμ-Tcl1 mice exhibited downregulated genes for proinflammatory mediators (52). Thus, TANs from Eμ-Tcl1 mice, which carry their tumor burden for several months, share properties of an N2 differentiation.

Moreover, we found enhanced gene expression for neutrophil-attractant factors in Eμ-Tcl1 stroma, an expansion of splenic Ly6G⁺ neutrophils, and an extended neutrophil lifespan. Neutrophils localized to the RP and MZ, a distribution that was clearly distinguishable from autoinflammatory processes where neutrophils can be located in the T-cell zone (53). Depletion of neutrophils in Eμ-Tcl1 mice delayed leukemia proliferation associated with the upregulation of the cell-cycle inhibitor Egr1 and the differentiation factor Egr2 (25, 26).

The functional properties of splenic neutrophils in Eμ-Tcl1 mice were reminiscent of N_BH cells important for MZ B-cell antibody production (27). An important cytokine for intrasplenic reprogramming of neutrophils into N_BH is IL10, which could be provided by leukemia B cells. We suggest that the local environment in leukemic spleens shapes the gene expression program of this N_BH-like subset, which does not exist in peripheral blood or in bone marrow. Because of the local priming, drawing parallels to human is tempting, but for ethical reasons splenic specimen from CLL patients are not available.

Immunosuppressive G-MDSCs share the surface markers CD11b, Ly6G, and Ly6C^low with neutrophils (29). Release of increased amounts of ROS was shown to mediate immunosuppression by G-MDSCs (54); however, in Eμ-Tcl1 neutrophils, ROS production was unaltered. Expression of two hallmark genes that are either strongly overexpressed in G-MDSCs versus neutrophils (Mpo), or vice versa downregulated (Ikbke) was not substantially different between Wt and Eμ-Tcl1 neutrophils. We conclude that splenic neutrophils in leukemic mice exhibit a stronger overlap with neutrophils than with G-MDSCs.

Gene signatures obtained from Eμ-Tcl1 splenic stroma reflect a homeostatic condition in which leukemia B-cell survival depends on the expansion and skewed differentiation of innate immune cells. Splenic neutrophils may act as tumor-instructed N_BH-like cells that provide the B-cell growth factors in excess, and thus influence leukemia B-cell ontogeny during B-cell selection and progression.

No potential conflicts of interest were disclosed.

Conception and design: R. Kettritz, G. Lenz, U.E. Höpken, A. Rehm

Development of methodology: M. Gätjen, M. Grau, K. Gerlach, J. Westermann, A. Rehm

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Gätjen, F. Brand, M. Grau, K. Gerlach, R. Kettritz, J. Westermann, I. Anagnostopoulos, U.E. Höpken, A. Rehm

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Gätjen, F. Brand, M. Grau, I. Anagnostopoulos, P. Lenz, G. Lenz, U.E. Höpken, A. Rehm

Writing, review, and/or revision of the manuscript: M. Gätjen, M. Grau, R. Kettritz, I. Anagnostopoulos, G. Lenz, U.E. Höpken, A. Rehm

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Gerlach, A. Rehm

Study supervision: U.E. Höpken, A. Rehm

We thank Kerstin Krüger (Max-Delbrück-Center, Berlin, Germany) for excellent technical assistance.

This work was supported by German Research Foundation (DFG) and in part by Wilhelm-Sander-Stiftung (A. Rehm and U.E. Höpken).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.